娇色导航

Keith Shaw Hi everybody.

Welcome to Global Tech Tales, where we talk with editors from around the world about the latest technology and leadership topics, and hear stories from IT leaders about what they're looking for. I'm Keith Shaw co hosting along with Matt Egan.

He is the global content and editorial director at Foundry, who's also represents the UK in these shows. And this month, we are joined by Jack gold, Principal Analyst at J Gold Associates. Welcome everybody. Good to see you. Jack. Thanks guys. All right.

And so for this episode, we are talking about how AI is transforming the data center. Last week or last month, we talked about how AI is disrupting cloud computing. And so this almost is like the opposite end of that scale for a lot of IT people.

So, you know, there are a lot of great reasons for running AI applications in your own data centers. AI is putting new demands on virtually every aspect of data centers, from servers, networks, power grids and more.

But beyond this decision about where to run AI workflows, other issues are looming in the background. So we're going to take a look at all of the different things, and we start the show off as always, with some statistics.

So this is what I found rummaging around the internet and including the 2025, state of the data center report from core site and Foundry, shows that the expansion of AI is pressuring organizations to reassess their IT infrastructure, to balance cost and performance with Co Location data centers now taking on an expanding role in the study.

This is an amazing number. 98% of IT leaders said they have adopted or plan to adopt a hybrid IT model. The research also suggests that cloud costs are driving organizations to repatriate apps and workloads from the cloud back to on premise data centers.

Another stat that blew my mind too was that the SMP global voice of the enterprise survey said that more than 70% of respondents are saying that their current IT infrastructure was inadequate for future machine learning and AI workloads. So that's down the road.

McKinsey said that 70% of total data center capacity demand will be aI related by 2030, and generative AI loan apps alone will be about 40% of that capacity.

And then data power, data center power demand is also expected to skyrocket, according to Goldman Sachs, power demand is going to rise 50% by 2027 and 165% by 2030 so you got to start thinking about all of your power issues.

And then finally, 2023 report by Uptime Institute said 58% of data center operators are struggling to find qualified candidates for vacancies, with shortages concentrated in junior and mid level operations. In addition, 34% of companies said they had no initiatives to initial to recruit and train new entrants.

So there's a lot of stats around the world of the data center these days.

So what I want to start off with, however, before we jump into the how AI is affecting things, before the show, we were talking about, the issue is that I think our perceptions of what a data center is these days is is just is a little off kilter.

So you know, when I think of data centers, I think of what my dad was doing back in the 70s where you had giant machines and punch cards and air conditioned rooms and things like that. But that perception is no longer the case, right Jack? Jack Gold

The common view of data centers, as you say, is this big, CPU, GPU, memory, tape factory is pretty much gone. In many cases, it's a relatively small room with racks that are pretty small compared to the old days, right?

So a typical computer these days is a, you know, a rack based system that plugs in. There's 10 or 12 or 14 or 18 of them in a small rack. That's, you know, I don't know, three feet by six feet tall, and that's your data center.

And that can be in, you know, in you know, in a closet, basically in a room.

Also, as we move to distributed compute, which is what data centers really are these days, you know, if you're aif you're a Walmart or Target or McDonald's, you've got data centers are distributed to all your different locations, so you've been at the edge.

So data center, the notion of what a data center is today is not this big room with lots of stuff in it. It's lots of stuff, potentially scattered all over the place. Keith Shaw

Yeah, is it fair to still use that definition, then, of data center as something that is unique to, you know, with the technology? Or should we just start calling it something else. Matt, like, you know, why? Why don't you wear it? Weigh in on this. Matt Egan

Keith, you know me, I like to be philosophical, you know, and I to think in a philosophical so I'm going to say, to a certain extent, and I'm being slightly facetious, but the data center is a state of mind, right? It's almost as a terminology.

It's almost become synonymous with infrastructure, right? Because Jack's exactly right. It's, it's, this is the physical space, but it's almost anonymous infrastructure. Now it's where your data is, it's how it's stored and accessed.

And Jack just spoke to the the infrastructure of these large organizations that's just kind of everywhere, right? For most organizations of any size, especially if they've been around for a while, it's likely to be a mix of those.

Physical spaces with servers, plus public or private cloud, maybe even some storage and computer at the edge.

I mean, the only thing that binds it all together, the one thing I will say is that, in the end, it is all about data stored on boxes, which is the piece that's, that's that's occasionally forgotten. But yeah, it's not, it's funny. You mentioned punch cards.

I'm immediately thinking of cable ties and, you know, you know, the neat cabling of the many, many data center tools I did in my data as a beat reporter kind of thing. But it, but it's not that now. It's a much more distributed concept. I think. Keith Shaw

Does that mean like infrastructure and data center then becomes a generic word for basically wherever a company is storing its data, and then how it's doing it. Matt Egan

I mean, I think in this instance, I'm using it that way, probably incorrectly. I mean, ultimately that physical space data center, Keith Shaw Yeah.

Okay, so last month, on the show, we talked about how AI was affecting cloud computing technologies, and then there were a lot of stories that came up where costs are spiraling out of control for many companies, and a lot of them are now reconsidering moving some of their data and workflows back to the more traditional data center, which might just be on premise somewhere, or a mix of these things, these co-location centers, or hybrid IT models.

So are you also seeing this play out in the real world? Jack. Jack Gold

Yes, for a couple of different reasons. Number one is that, originally, the reason why most companies went to AWS, Azure, Google Cloud, Oracle, IBM, pick your favorite, right, is that they needed extra capacity, and they thought that a OPEX model versus a CapEx model would be more effective.

It's a turn as it turns out, as and it was early on, as it turns out, as prices went up by most the hyperscalers, it wasn't as clear.

Number one, number two is that even if you're not using all the compute that you're you're scheduled to use at these hyperscalers, they're still charging you to keep that data on prem, for them on their Prem, not on your Prem, right? And so there's a cost involved.

Number three is that people are finding that from a privacy issue, from a security issue, and more importantly, these days, from a data sovereignty issue, that keeping data local to my systems gives me a lot of advantage, advantages, not just in cost, but in also an operational efficiencies.

And so we're seeing a lot more companies evaluate their cloud infrastructure, if you will, their or their hyperscaler infrastructure, where they're going with their data and their compute and looking at it and saying, Does it really make sense for us to go there just because they have it available, for sure, they have the compute available that we may not have, but just because we may not need all of that, and it may be more cost effective to keep it on board.

Matt Egan

Not to relitigate old ground. But we did. We talked about this last month in the context of cloud, right, the sense that the one thing new about cloud is wrong.

We were always told, as Jack, just articulated, that centralizing in public crowd would be more efficient reduce cost and overhead, but the complexity of the nature of every organization means that some combination of the nature of business and capitalism and the complex needs of the data layer, especially in the age of AI, have really shifted that mindset.

I don't speak with many, if any, IT leaders who see their future all in the cloud, right? There's some element of data center and some combination of cost, privacy and legislation that needs to be flexible. Means having a data center of one's own is really important. Keith Shaw

Now, if there are some companies that are considering like, they see these cloud costs, and they see a lot of these other issues, and they're going, oh, man, we have to get our data back to the data center.

It's not going to be smooth sailing for a lot of these, these companies as well, right? There are some, some hidden obstacles or hidden costs that might be involved, right? Jack. Jack Gold

There are a lot of costs.

But I just want to make another point that on top of Matt's, and that is that even the hyperscalers, even cloud providers, understand that it's going to be distributed, and they all have remote clouds, for lack of a better term, where they'll drop a server at your location and just manage it from their cloud so it they know that this is coming.

They know that they have to modify this now from a cost perspective, yeah, it's not all that necessarily cheap to go local, right?

You need to have compute, you need to have storage, you need to have connectivity, you need to have management of the system, so it needs to be involved and so, and by the way, if you're going to repatriate your data, the cloud guys are going to charge you for that.

It's not free. Even though it's your data, it's not they hold it ransom, basically, if you want it back. And so that may maybe that's a little unfair, but it's but that's really what happens.

And so companies look at this and say, okay, but it's a one time cost, right? The companies I talked to have said, we're doing evaluation, right? It's all about TCO, it's all about ROI, it's all about dollars at the end of the day.

And we look at this and we say, okay, in the short term, it's going to cost us X. In the long term, it's gonna save us. Why? And so that's you're going this route. Keith Shaw

And but then another problem is, is that we don't know if a lot of these companies would have the infrastructure to handle a lot of these workflows that are moving back.

Like, let's say that you have concerns about AI workflows from a security and a data privacy standpoint, and you're like, Okay, I'll move it back to the data center, but that will impact, impact the capacity, right? And workloads on those, those on premises servers, right?

Jack Gold Yeah, for sure.

Keith Shaw

Do you think it leaders are thinking about that, or do they already? You think that they know about that already? Matt Egan

The reality is, every organization is trying to build its, its its future tech stack to deal with what's coming next, right?

And that that ranges everything from for a lot of lot of organizations, it may just make sense to be signed up with a large cloud provider who's promising always to have next gen technology, and if you're only looking to drive efficiencies for kind of inside AI, that might be right.

It might be right. But I think for most organizations, it's more complex than that, and so we're not talking about one size fits all solution, even at the individual user level.

I know we'll talk about aipcs and distributed edge computing, but it's not about every single employee having access to the same things. It's about understanding what your needs are and providing the best combination of solutions and technologies to support that. Jack Gold

There's another component to that, and that is that it really depends on who's driving the initiative right.

In some cases, I've seen companies it's being driven top down, basically by edict the C, whatever, O, will look at this and say, hey, it's too costly do this, as opposed to it saying, This is what we need.

And so it becomes an issue of, where is the expertise, what's driving, who's driving? That that need within the organization, companies, from a kind of a top down. Now the top down. I shouldn't, I shouldn't dismiss it. They have to be involved. Obviously, they've got the money.

They control the money, right? They're signing the checks, right? But if it's being done by edict, right, it doesn't work very well. And so the organization needs to understand the overall organization.

Needs to understand that you got to have a balance between the guys driving it from the top down and the guys driving it from the bottom up, who are the technologists, who knows what's needed, and there's a lot of disconnect in many organizations between those two. Keith Shaw

So does it look like that? We're heading towards a world where it's a hybrid model of some stuff on premise, some stuff in Co Location centers, some stuff in the cloud, like, that's where we're all moving, right?

Similar to the whole remote work at home types of things too, right? Matt Egan

Yeah, I think so. And I actually think you did a great job with the stats at the beginning, Keith, I never give you enough praise because, because, because they did literally speak to that, right?

They literally, there was, there was a couple of stats in there, literally said that this is the future. Every organization has its own different set of needs.

Within every organization, we tend to talk about it's a little bit like Jack was just saying the top down edict versus the operational approach. We tend to talk about organizations as if they are just one process or one function.

And like within organizations, you've got wildly different needs for access to data and the technology that supports that. And so I can't think of a single organization that I have conversations with where it is, like, one, like, kind of generic solution for everything.

So I think that is definitely the world we're headed to. It's some combination of, and I think, and I think it's three factors. It is public, private cloud. It is on premise data center, and it is some level of distributed at the user device level compute. Keith Shaw

These issues were probably being discussed in, you know, by IT leaders for many years. You know, obviously before generative AI started coming around.

But I think again, we're looking at this with the lens of AI, and is AI now forcing companies to do something that they might not have wanted to do before because of the the amount of workloads and capacity and power demands, things like that. Jack Gold Yeah.

So, so everyone is looking at AI, everyone says that I've got to be involved with AI. It's coming. It's, you know, it's going to hit me over the head if I if I just ignore it that that doesn't work.

On the other hand, probably 60, 70% of companies that are already doing something with AI have not really found a good reason to deploy it widely within the organization. And so it becomes an issue of, where is AI good for my organization, and how do I deploy it?

The other issue for many organizations now is, and this is also related to what Matt was just saying, is that, from a distributed perspective, costs have come down.

You know, I can buy for under 10k a pretty significant equivalent of a data center that you know, from just two or three years ago. And and put it, you know, it's a small box.

It's the size of, you know, a small refrigerator, bit maybe, and you can put it over in a corner and plug it in, and it'll do all kinds of stuff for us with AI, as AI is coming, we're going to see even more of that with our estimate is that if you look out two to three years, 80% of AI workloads are going to be running as inference workloads on remote devices.

That right remote device might be anything from a, you know, a small server, to AI PCs, which we, you know, are coming as well to your phone. I mean, look at, look at the power you got on your on your mobile device these days.

And so companies need to start thinking about how they're going to make that a distributed environment, from an AI perspective. And it gets really complex really quickly, yeah. Matt Egan

And I think also like it again, we're talking about as if all organizations the same, and that's that's to do with what the organization does, like who their customers are, what their goal what their goals are.

It's also to do with their level of maturity around that particular subject, right? Because, because because I completely hear what you're saying, Jack, I hear all the time from IT leaders right around this idea of AI is here we've got to do something kind of thing.

But it does. It does range from the we're going to work with our existing SaaS vendors and use their capabilities, and that's just going to make things happen a bit faster. Somehow, we're going to reduce some toil somewhere.

So we're going to completely rethink our customer journey and our products and solutions, and therefore the way we manufacture them, physical or otherwise.

And therefore, again, like, some combination of those solutions, like, is going to be it's going to be required to support like, from an infrastructure perspective, that those outcomes and what's really interesting about what you're just talking about what you're just talking about there Jack the distributed model, the fact you're absolutely right, and I didn't even consider that like, right down to the server level, how much less expensive it is now, from a CapEx perspective, to go out and get these things.

That is, is this idea of scalability, of being able to plug in little elements here, now and everywhere, geographically, where they're needed, from a function perspective, where they're needed that is that is going to be a big part of like the adaptation process over the next little while.

I would say. Keith Shaw

Do you think that you're going to be seeing CFOs, that are going to be okay with a lot of that OPEX money moving back into capex, or is it because the cap ex equipment and the hardware is less expensive that it wouldn't be as big of a problem as it might have been before.

Jack Gold

I think you're going to see them looking at it from the perspective of money, whether it's OPEX or capex, doesn't matter. What's it going to cost me to run this workload if I can do it cheaper on my own box, in my own facility, great.

If it's cheaper to run it at, you know, AWS or Azure, okay, do it there. It's going to be really a cost justification across the board.

And I think when we we lose sight of the fact that money is money, whether it's OPEX or capex, it's still coming out of the bottom line of the company. That's what happens when people you know get confused about where do I need to go?

And I think that a lot of that happens in it today, especially since it in the past, really has not been run kind of as a profit center. It's really been run as a thought center, right?

And, you know, I get a budget, I blow it, I, you know, I go through it, it doesn't, it isn't, doesn't roll into the overall finances of the company. And I think that's changing dramatically in a lot of different businesses. Matt Egan

I think it is Jack. I think you may also speak to some very successful, well run organizations, right? Because, I because, because I also say, Well, there's a there's a load of factors in here, right? Your point about it is, is absolutely correct.

And in some organizations, like I was saying before, like where products and services have been rethought. Customer journeys have been rethought using AI and the and the IT strategy is the strategy. Then you're absolutely right.

It becomes like a profit center, but also, like it depends on the length of horizon for an organization, right? If you're, if you're owned by a PE firm, or like you've got shareholders, and like it's living quarter to quarter, then actually, OPEX can be quite attractive.

CapEx can be really attractive because you can chuck costs in there, and they don't really count against the balance sheet kind of thing.

But without wishing to be cynical, you know, if you're, if you're a company that's owned by Pe and you're looking to be sold or exit, then then, you know, burying costs in capex can be quite a good thing.

So what I'm saying is, I completely and 100% endorse everything Jack said, but not all organizations are run in an optimal manner, and so you do still see this kind of chaos around, like it as a cost center and short term decisions that might not money is money, but in the end, like when you need to report that money is a big factor for lots.

Jack Gold

It also depends on the age of The organization. Matt, if the company is three years old, you know, you're a startup, you're going to run very differently than a company that's been in business for 30 years or 80 years, or 120 years, right? Yeah.

I mean, they're, they're guys out there, and I won't name companies, but I've talked to them where, you know, they're still, still running Windows 95 on some of their systems. I mean, you know, you got on an airplane, they'll start.

They're surrounding dot matrix printers to print off the list. So you're right. Matt, it varies all over the place. Keith Shaw

Is this an example where the lot of the older companies might have an advantage of of moving data back to a data center, whether it's a hybrid cloud or, you know, something other, other than just, you know, because they've had that experience in data centers before.

Or Or are the startups more likely to be like, Oh, we've been flexible before. We can be flexible by, you know, hosting things in our own data center, and just more about the size of the company or the practices you had a gut feeling on any of those.

Jack Gold

Keith, the problem with some of the companies have been around for a long time is their data centers are outdated.

You know, they didn't the reason they went to the cloud is because they wanted better compute, better memory, better interconnectivity, and so they haven't really spent a lot of money on upgrading their data centers.

Now, if they want to move this stuff back and they want to get an equivalent level of performance. They have to upgrade those data centers. They're willing to do that now because the costs are less. But it's not just an issue of costs.

It's also an issue of skill sets. It's also an issue of capabilities within the organization to be able to do that.

And a lot of organizations, as you know, you said in the statistics, a lot of IT organizations don't have the resources necessary to pull off a lot of the stuff that management wants them to do. Keith Shaw

I was going to make a joke about about maybe companies do it running their AI workloads on mainframes. But I wouldn't be surprised if some of those companies actually did that. At some point. Jack Gold

They do that's, mainframes are not dead. Keith Shaw Yeah.

Basically, you know, my entire IT career has been talking about whether mainframes are finally gone or not. They never are.

I want to raise another question that we that we brought up in the stats, which was the last one the report that said 58% of data center operators are struggling to find qualified candidates.

In addition, 34% of companies said that they don't even have any new initiatives to recruit and train data center workers.

Is that something that maybe people aren't aware of yet either, is that if you are going to be moving a lot of this data back to a data center on premise that that's going to raise some issues of finding the right people, right?

It's, it's a huge issue for a lot of these companies. Matt Egan

It's a huge problem and and I think, I think it is a very acknowledged problem, to be honest, people, when I speak to it, leaders, it's something they're very aware of, albeit potentially the specific instance around the data center may not be the first thing that comes to mind.

I mean, certainly when we talk about the skills gap like it is, across the board, there's a huge skills gap, even as there are a myriad IT pros out of work.

And there's no doubt that lack of skilled workers in data center management is going to be an issue.

There's also an issue like I say, I think it leaders tell me that they recognize, but potentially business leaders are less sensitive to because, because what I do hear a little bit from IT leaders is that there's an expectation that you can throw money at that particular problem and get it resolved.

And that's not necessarily that simple. But Jack, I'm really interested to hear what you hear from from you. Jack Gold

I hear the same thing, Matt, it's, it's it. It depends on who's driving the initiatives, the upgrades, right? CEOs, CFOs, you know, the C suite. Look at this and say, Yeah, let's throw money at it. You can just go hire more people.

The problem is, those people aren't out there. I don't remember the statistics anymore, but I've seen reports where they said, you know, within the next 10 years, there's going to be a shortage of, you know, 5 million, it people. They're they're not, they're just not gonna be available.

And frankly, it's going to get worse.

Because if you see what's going on, I don't mean to turn this into political statement, but if you see what's going on in this country, you know, if you can't get immigrants in that are filling a lot of the jobs, it becomes even worse. Matt Egan

It's the same in my country, like exactly the same and the same, the same drivers driving it, which, which is intriguing from a political perspective, but let's just talk about economics and it right, it's a real problem. Keith Shaw

I mean, the other issue too is that, is it the fact that maybe data center skills are not as as exciting as maybe some of these new skills?

You know, we write reports all the time about the hottest job skills and the, you know, the jobs of the future, it's all going to be aI related.

You know, where does data center, you know, data center operations, junior level, mid level engineers like that, probably doesn't come at the top of the list, right? Jack Gold

Well, if you think it's bad now, wait till we have aI needs. Where are you going to get the skills for those guys? Matt Egan

We have folks right now.

I can see them doing some building work on my house and, you know, and whilst he was digging a huge ditch that I couldn't possibly do, you know, this guy was saying to me the other day, oh, you know, I wish I could do what you do.

And I was like, Dude, I wish I could do what you do, because, like, like, you're always going to be needed, right?

And, like I said before, at the end of the day, all of this stuff boils down to the to the boxes, like you need the compute, you need the storage, you need the connectivity, you need the people who the engineers, is right.

You need the people who can make that work. Or none of this other stuff happens.

So it may not always be the most kind of celebrated aspects of this work, but, but it's, it's to Jack's point is it's really important, and it's likely that that need is going to get more apparent. I would say. Keith Shaw

I want to bring up a one other topic around this, this idea, and that is, we had mentioned the move of a lot of companies that are bringing out AI PCs. Ai chips are getting on the on the phone and Jack.

I know you covered the chip maker market as well. Ai chips are going into servers and going into a lot of these things that will be running in the data center. What kind of impact are they having on the market?

Is that, even driving this, this move from the cloud in, you know, almost to the edge, is that, is that driving, you know, the space Jack Gold

Absolutely, you're going to see AI running at the edge more and more, whether it's an aipc, whether it's your phone, whether it's your car. You know, in a couple of years, you'll you'll have, you'll be driving a data center around when you have a car.

And so a lot of the applications that we're talking about, a lot of the infrastructure, a lot of the interconnect requirements, are going to move to the edge. You're going to see a huge move in that direction. Aipcs are already having an effect.

Because we're seeing companies looking at acquiring a but not so much the AI PCs that you buy for a general worker, but those that are workstation level that we put out, we're already seeing AI workloads move down to those because as as companies experiment as companies.

Have AI engineers that need to do stuff. Why do that in the cloud when you can do that on your own box at PC level?

Now, it doesn't have the same level of compute, but generally you're not doing, you know, huge training of models that have 30 billion parameters, or three 300 billion parameters, or three Tera parameters, or whatever the next one is, and so we're already seeing that effect come on board.

It's still slow. It's not going to take over overnight, but over the next couple of years, you're going to see those things proliferate in most organizations.

So absolutely, it's going to become, by the way, from the IC or the processor perspective, no one would consider buying an integrated processor today that didn't have a GPU that wasn't the case 10 years ago, right?

The same is going to be true in the next couple of years with NPUs AI acceleration, you're never going to buy a processor without that built in. Keith Shaw

So I did talk to, I have talked to some people, and they they're a little bit skeptical on the the aipc side of things.

But you know when, when you talk with IT leaders, are they okay with this, this, this movement towards the edge, or is it, or is it just a lot of hype from some of the vendors because they need to sell hardware? Matt Egan

I think it's both, honestly, like, I think, I think the way that aipcs have been marketing around the end of Windows 10 support, like, as if large enterprises are going to replace their entire fleet with significantly more expensive, you know, laptops for general workers.

And then that is hype, honestly. And there's a bit of politics involved there, from a business level, you know, with partnerships between vendors, chip makers, the OEMs like.

But what I would say, and I've changed my views on this quite a lot over the past couple of months, talking to analysts and IT leaders around the use cases for the exactly what Jack's talking about the edge computing workstation level. That's a brilliant way of putting it.

I hadn't thought of it like that before, but I was speaking to an analyst recently in an event that was talking about, not everybody's going to have an aipc, right? That's not what it's going to be. Jack's right?

You're going to buy your PC is going to come with an NPU.

That's just table stakes at this stage, in exactly the same way as the graphics chip, but like not everyone's going to have have a need for this all bells and whistles mega computer, but for specific functions that are working with data and analytics at that AI level of compute.

It makes total sense to have clusters of people almost creating mini data centers between their AI PCs at the edge using these devices, and then when you think about that, that takes away some of the issues around security and data sovereignty and scalability you might have if you put in things into a public cloud.

So that I'm starting to hear that become more of an accepted use case, rather than these people are trying to sell me AI PCs. They cost a lot more than the PC I bought before. I can't afford to refresh the fleet.

It's more about we're starting to figure out what our AI strategy is, and our IT strategy underpinning that will involve some number of our workers using AI PCs in order to create distributed computing to support those activities and those specific roles and functions. Keith Shaw

Jack, what are, what are the IT leaders talking to you about? Are they? Are they on board with this, or are they skeptical? And then, and then your own opinion too, if you want to, if you want to throw that in there. Jack Gold

Sure, I have an opinion. Wow. Matt Egan

I'm never allowed an opinion. Jack Gold Yeah, exactly.

Look, I think they're cautiously optimistic that this thing is all going to play out. Ai PCs today are mostly about productivity tools, right? It's, it's it's office 365, it's what Microsoft is putting on, it's co pilot kind of stuff.

But longer term, what they're looking at is looking at AI PCs that are going to be handed out, handed out or. Are made available to people who are doing service calls. Are being handed out to aipcs.

Are going to go into McDonald's, and as you order your your, you know, your burger, they'll tell you how to cook it, or what, what.

I don't know how long you're gonna have to wait before the fries are done, or, you know, whatever, whatever's gonna happen in that space. They're looking at that from the perspective of what can I do with it in the future?

As far as buying AI PCs, most companies, when they refresh, you know, they refresh their PCs, their laptops, every three years or so, give or take.

And if you look at what they usually buy, they usually buy a little bit up from, you know, the base level model, just because they know they're gonna have it for three years. They don't know what's gonna with for gonna with processing is gonna look like.

They know they're gonna need more thing in three years. That's just a given. And so there is a lot of looking at, okay, I need to buy an aipc Because I, you know, I know stuff is coming. I'm gonna buy it, and it's there is a cost uplift.

Depends on the on the machine.

It could be a couple $100 it could be 567, $100 or $1,000 567, $100 or $1,000 depends on the machine again, but they're looking at doing that, but, but ultimately, long term companies know that they're going to need to buy AI enabled devices because it's coming.

You can't avoid it. I mean, it's like, like the web was right? I remember, I'm an old guy. I remember when people were saying, I'll never need emails. What do I need email? You know, I'll never need a web browser. What good is that to me?

You know, the heck with it. Keith Shaw

I'd be fine if email went away. Jack, to be honest. Okay with that, not all email, 90% of it, 90% of it, yeah, all right. So I want to, you know, we wrap up the show, usually with the the end of show vote.

And, you know, sometimes it goes well, most of the times it's always, it's always a yes, no, maybe. So I'm going to try to see if we can get a yes or no answer, or at least, you know, two choices.

But you're always free to just say, it depends. So, you know, here's the question, are problems in the data center a major concern or a minor concern for IT leaders in 2025. Jack Gold It depends.

Keith Shaw

I knew you were gonna say that Jack Gold

honestly, Keith, there isn't one answer for everyone, right? We speak in generalities on the show, and I completely understand, but it is, for sure, a continuing problem that's gonna be around for a very long time. Matt Egan

Yeah, I mean, I, I'm sorry, Keith, I've got to do it. I don't accept the premise of the question. I should think it's about opportunities. There are problems to be solved, right?

But I think it's about opportunities, because if you get your data structure right, not perfect, but right, and then you find the best combination of data center, cloud, distributed end user, device, like, actually, that's, that's the opportunity.

But, yeah, that is a problem to be solved for all IT leaders and it and it's a major concern in that sense. Keith Shaw

Yeah, okay, I think it's, it's not a major concern. It is one of those things on the checklist that that most companies have probably already been discussing.

AI adds a little bit of a new wrinkle to it, but I'm pretty confident that most companies, if they're prepared, and if they, you know, watch shows like ours and and are ready for it will be, you know, ready for this as well.

So I don't you know, it's always the companies that are probably lagging behind that where you're going to start seeing a lot of the hiccups. So, and Matt, I'm going to have you start writing the yes or no questions for future episodes? Okay, that seems reasonable.

That's all the time we have for the show today. Thanks for everybody for participating. Jack, always a pleasure talking with you. Jack Gold

Thank you, Keith, great being here. Keith Shaw

All right, and we're gonna be back next month with more global editors talking about how IT leaders are managing the AI disruption. So feel free to add any comments you have below.

Be sure to check out our other tech talk shows, such as today in tech, cio leadership live and demo if you are interested in seeing B to B product demonstrations. I'm Keith Shaw, thanks for watching. Transcribed by https://otter.ai